Left-to-right processes in the construction of sentences

The purpose of this study was to test certain predictions derived from an extension of Yngve's left-to-right model of sentence construction. On the basis of assumed development of predictive integrations between successive rule-applications, it was predicted that for a right-branching language such as English, it should be more difficult to provide the beginning of an already completed sentence, than to provide the end of an already started sentence. But for a left-branching language such as Turkish, this tendency should be considerably reduced. The obtained results supported these predictions. It was also found for English that this result could not have been solely due to differential decoding times.